Downloads · 30 days
12
32% of all-time downloads
donut_experiment_bayesian_trial_0
davelotito/donut_experiment_bayesian_trial_0
donut_experiment_bayesian_trial_0 is a image-text-to-text model from davelotito. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as mit.
should probably proofread and complete it, then remove this comment. --
All-time downloads
37
Public
Parameters
202M
6.5 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors809 MB · 99%
Parameter types
How the weights are stored.
F32202M · 100%
Try a prompt
Base models
Model card
From the Hugging Face model README
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
donut_experiment_bayesian_trial_0
This model is a fine-tuned version of naver-clova-ix/donut-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4050
- Bleu: 0.0639
- Precisions: [0.79957805907173, 0.7386091127098321, 0.7083333333333334, 0.6765676567656765]
- Brevity Penalty: 0.0876
- Length Ratio: 0.2912
- Translation Length: 474
- Reference Length: 1628
- Cer: 0.7653
- Wer: 0.8371
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1.2045081648781836e-05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 2
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Precisions | Brevity Penalty | Length Ratio | Translation Length | Reference Length | Cer | Wer |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.9353 | 1.0 | 253 | 0.6228 | 0.0486 | [0.7096774193548387, 0.6053921568627451, 0.5612535612535613, 0.5102040816326531] | 0.0820 | 0.2856 | 465 | 1628 | 0.7751 | 0.8592 |
| 0.462 | 2.0 | 506 | 0.4846 | 0.0568 | [0.7913978494623656, 0.7058823529411765, 0.6609686609686609, 0.6224489795918368] | 0.0820 | 0.2856 | 465 | 1628 | 0.7650 | 0.8423 |
| 0.4071 | 3.0 | 759 | 0.4226 | 0.0626 | [0.7899159663865546, 0.711217183770883, 0.6767955801104972, 0.6459016393442623] | 0.0889 | 0.2924 | 476 | 1628 | 0.7685 | 0.8436 |
| 0.3007 | 4.0 | 1012 | 0.4092 | 0.0638 | [0.7957894736842105, 0.7344497607655502, 0.7008310249307479, 0.6644736842105263] | 0.0883 | 0.2918 | 475 | 1628 | 0.7640 | 0.8397 |
| 0.3114 | 5.0 | 1265 | 0.4050 | 0.0639 | [0.79957805907173, 0.7386091127098321, 0.7083333333333334, 0.6765676567656765] | 0.0876 | 0.2912 | 474 | 1628 | 0.7653 | 0.8371 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.1.0
- Datasets 2.18.0
- Tokenizers 0.19.1
- Task
- Image-Text-to-Text
- Library
- transformers
- Type
- vision-encoder-decoder
- License
- mit
- Created
- May 20, 2024
- Updated
- May 30, 2024